Los Alamos National Laboratory
Increasingly, scholarly artifacts created throughout the research lifecycle such as source code, datasets, audio/visual recordings, technical reports, presentation slides, and blog posts are considered necessary components of the scholarly record. Our previous research confirmed this claim and revealed the widespread use of version control systems among academic researchers writing source code. However, we also noticed a lack of comprehensive institutional programs to preserve their scholars’ source code. In addition, there are very few technical systems and tools available for the high-quality capture of source code artifacts at scale.
In response to this observation, and with generous support from the Alfred P. Sloan Foundation, we started the “Collaborative Software Archiving for Institutions (CoSAI)” project as a collaboration between New York University, Los Alamos National Laboratory, the University of Pittsburgh, and Old Dominion University. The CoSAI project focuses on developing and evaluating institutional approaches to archive research software hosted on public code sharing platforms on the web. This includes the creation (and sharing) of institutional collection policy plans as well as the technical development of systems that do not just crawl the code hosted on sites like GitHub, but also reliably capture the ephemeral record of supplementary material related to the code, for example, discussion threads and issues. We consider these materials just as important as the code itself, specifically for the purpose of reproducibility and replicability. By leveraging existing open source tools like Memento Tracer and building on workflow engines such as OCCAM, CoSAI will contribute to capturing web resources from code repositories at high quality, at scale, and in a reproducible manner. The briefing will include a project plan outline and report on progress to date.
See also “Who Versions Scholarly Code” (https://youtu.be/hmhMaEGQA6o) from CNI’s Pre-Recorded Project Briefing Series, Oct. ’22.